path: root/erts/doc/src/absform.xml

<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">

<chapter>
  <header>
    <copyright>
      <year>2001</year><year>2011</year>
      <holder>Ericsson AB. All Rights Reserved.</holder>
    </copyright>
    <legalnotice>
      The contents of this file are subject to the Erlang Public License,
      Version 1.1, (the "License"); you may not use this file except in
      compliance with the License. You should have received a copy of the
      Erlang Public License along with this software. If not, it can be
      retrieved online at http://www.erlang.org/.
    
      Software distributed under the License is distributed on an "AS IS"
      basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
      the License for the specific language governing rights and limitations
      under the License.
    
    </legalnotice>

    <title>The Abstract Format</title>
    <prepared>Arndt Jonasson</prepared>
    <responsible>Kenneth Lundin</responsible>
    <docno>1</docno>
    <approved>Jultomten</approved>
    <checked></checked>
    <date>00-12-01</date>
    <rev>A</rev>
    <file>absform.xml</file>
  </header>
  <p></p>
  <p>This document describes the standard representation of parse trees for Erlang
    programs as Erlang terms. This representation is known as the <em>abstract format</em>.
    Functions dealing with such parse trees are <c><![CDATA[compile:forms/[1,2]]]></c>
    and functions in the modules
    <c><![CDATA[epp]]></c>,
    <c><![CDATA[erl_eval]]></c>,
    <c><![CDATA[erl_lint]]></c>,
    <c><![CDATA[erl_pp]]></c>,
    <c><![CDATA[erl_parse]]></c>,
    and
    <c><![CDATA[io]]></c>.
    They are also used as input and output for parse transforms (see the module
    <c><![CDATA[compile]]></c>).</p>
  <p>We use the function <c><![CDATA[Rep]]></c> to denote the mapping from an Erlang source
    construct <c><![CDATA[C]]></c> to its abstract format representation <c><![CDATA[R]]></c>, and write
    <c><![CDATA[R = Rep(C)]]></c>.
    </p>
  <p>The word <c><![CDATA[LINE]]></c> below represents an integer, and denotes the
    number of the line in the source file where the construction occurred.
    Several instances of <c><![CDATA[LINE]]></c> in the same construction may denote
    different lines.</p>
  <p>Since operators are not terms in their own right, when operators are
    mentioned below, the representation of an operator should be taken to
    be the atom with a printname consisting of the same characters as the
    operator.
    </p>

  <section>
    <title>Module declarations and forms</title>
    <p>A module declaration consists of a sequence of forms that are either
      function declarations or attributes.</p>
    <list type="bulleted">
      <item>If D is a module declaration consisting of the forms
      <c><![CDATA[F_1]]></c>, ..., <c><![CDATA[F_k]]></c>, then
       Rep(D) = <c><![CDATA[[Rep(F_1), ..., Rep(F_k)]]]></c>.</item>
      <item>If F is an attribute <c><![CDATA[-module(Mod)]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,module,Mod}]]></c>.</item>
      <item>If F is an attribute <c><![CDATA[-export([Fun_1/A_1, ..., Fun_k/A_k])]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,export,[{Fun_1,A_1}, ..., {Fun_k,A_k}]}]]></c>.</item>
      <item>If F is an attribute <c><![CDATA[-import(Mod,[Fun_1/A_1, ..., Fun_k/A_k])]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,import,{Mod,[{Fun_1,A_1}, ..., {Fun_k,A_k}]}}]]></c>.</item>
      <item>If F is an attribute <c><![CDATA[-compile(Options)]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,compile,Options}]]></c>.</item>
      <item>If F is an attribute <c><![CDATA[-file(File,Line)]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,file,{File,Line}}]]></c>.</item>
      <item>If F is a record declaration <c><![CDATA[-record(Name,{V_1, ..., V_k})]]></c>, then
       Rep(F) =
      <c><![CDATA[{attribute,LINE,record,{Name,[Rep(V_1), ..., Rep(V_k)]}}]]></c>. For Rep(V), see below.</item>
      <item>If F is a wild attribute <c><![CDATA[-A(T)]]></c>, then
       Rep(F) = <c><![CDATA[{attribute,LINE,A,T}]]></c>.
      <br></br></item>
      <item>If F is a function declaration <c><![CDATA[Name Fc_1 ; ... ; Name Fc_k]]></c>, 
       where each <c><![CDATA[Fc_i]]></c> is a function clause with a
       pattern sequence of the same length <c><![CDATA[Arity]]></c>, then
       Rep(F) = <c><![CDATA[{function,LINE,Name,Arity,[Rep(Fc_1), ...,Rep(Fc_k)]}]]></c>.</item>
    </list>

    <section>
      <title>Record fields</title>
      <p>Each field in a record declaration may have an optional
        explicit default initializer expression</p>
      <list type="bulleted">
        <item>If V is <c><![CDATA[A]]></c>, then
         Rep(V) = <c><![CDATA[{record_field,LINE,Rep(A)}]]></c>.</item>
        <item>If V is <c><![CDATA[A = E]]></c>, then
         Rep(V) = <c><![CDATA[{record_field,LINE,Rep(A),Rep(E)}]]></c>.</item>
      </list>
    </section>

    <section>
      <title>Representation of parse errors and end of file</title>
      <p>In addition to the representations of forms, the list that represents
        a module declaration (as returned by functions in <c><![CDATA[erl_parse]]></c> and
        <c><![CDATA[epp]]></c>) may contain tuples <c><![CDATA[{error,E}]]></c> and <c><![CDATA[{warning,W}]]></c>, denoting
        syntactically incorrect forms and warnings, and <c><![CDATA[{eof,LINE}]]></c>, denoting an end
        of stream encountered before a complete form had been parsed.</p>
    </section>
  </section>

  <section>
    <title>Atomic literals</title>
    <p>There are five kinds of atomic literals, which are represented in the
      same way in patterns, expressions and guards:</p>
    <list type="bulleted">
      <item>If L is an integer or character literal, then
       Rep(L) = <c><![CDATA[{integer,LINE,L}]]></c>.</item>
      <item>If L is a float literal, then
       Rep(L) = <c><![CDATA[{float,LINE,L}]]></c>.</item>
      <item>If L is a string literal consisting of the characters
      <c><![CDATA[C_1]]></c>, ..., <c><![CDATA[C_k]]></c>, then
       Rep(L) = <c><![CDATA[{string,LINE,[C_1, ..., C_k]}]]></c>.</item>
      <item>If L is an atom literal, then
       Rep(L) = <c><![CDATA[{atom,LINE,L}]]></c>.</item>
    </list>
    <p>Note that negative integer and float literals do not occur as such; they are
      parsed as an application of the unary negation operator.</p>
  </section>

  <section>
    <title>Patterns</title>
    <p>If <c><![CDATA[Ps]]></c> is a sequence of patterns <c><![CDATA[P_1, ..., P_k]]></c>, then
      Rep(Ps) = <c><![CDATA[[Rep(P_1), ..., Rep(P_k)]]]></c>. Such sequences occur as the
      list of arguments to a function or fun.</p>
    <p>Individual patterns are represented as follows:</p>
    <list type="bulleted">
      <item>If P is an atomic literal L, then Rep(P) = Rep(L).</item>
      <item>If P is a compound pattern <c><![CDATA[P_1 = P_2]]></c>, then
       Rep(P) = <c><![CDATA[{match,LINE,Rep(P_1),Rep(P_2)}]]></c>.</item>
      <item>If P is a variable pattern <c><![CDATA[V]]></c>, then
       Rep(P) = <c><![CDATA[{var,LINE,A}]]></c>,
       where A is an atom with a printname consisting of the same characters as
      <c><![CDATA[V]]></c>.</item>
      <item>If P is a universal pattern <c><![CDATA[_]]></c>, then
       Rep(P) = <c><![CDATA[{var,LINE,'_'}]]></c>.</item>
      <item>If P is a tuple pattern <c><![CDATA[{P_1, ..., P_k}]]></c>, then
       Rep(P) = <c><![CDATA[{tuple,LINE,[Rep(P_1), ..., Rep(P_k)]}]]></c>.</item>
      <item>If P is a nil pattern <c><![CDATA[[]]]></c>, then
       Rep(P) = <c><![CDATA[{nil,LINE}]]></c>.</item>
      <item>If P is a cons pattern <c><![CDATA[[P_h | P_t]]]></c>, then
       Rep(P) = <c><![CDATA[{cons,LINE,Rep(P_h),Rep(P_t)}]]></c>.</item>
      <item>If E is a binary pattern <c><![CDATA[<<P_1:Size_1/TSL_1, ..., P_k:Size_k/TSL_k>>]]></c>, then
       Rep(E) = <c><![CDATA[{bin,LINE,[{bin_element,LINE,Rep(P_1),Rep(Size_1),Rep(TSL_1)}, ..., {bin_element,LINE,Rep(P_k),Rep(Size_k),Rep(TSL_k)}]}]]></c>.
       For Rep(TSL), see below.
       An omitted <c><![CDATA[Size]]></c> is represented by <c><![CDATA[default]]></c>. An omitted <c><![CDATA[TSL]]></c>
       (type specifier list) is represented by <c><![CDATA[default]]></c>.</item>
      <item>If P is <c><![CDATA[P_1 Op P_2]]></c>, where <c><![CDATA[Op]]></c> is a binary operator (this
       is either an occurrence of <c><![CDATA[++]]></c> applied to a literal string or character
       list, or an occurrence of an expression that can be evaluated to a number
       at compile time),
       then Rep(P) = <c><![CDATA[{op,LINE,Op,Rep(P_1),Rep(P_2)}]]></c>.</item>
      <item>If P is <c><![CDATA[Op P_0]]></c>, where <c><![CDATA[Op]]></c> is a unary operator (this is an
       occurrence of an expression that can be evaluated to a number at compile
       time), then Rep(P) = <c><![CDATA[{op,LINE,Op,Rep(P_0)}]]></c>.</item>
      <item>If P is a record pattern <c><![CDATA[#Name{Field_1=P_1, ..., Field_k=P_k}]]></c>,
       then Rep(P) =
      <c><![CDATA[{record,LINE,Name, [{record_field,LINE,Rep(Field_1),Rep(P_1)}, ..., {record_field,LINE,Rep(Field_k),Rep(P_k)}]}]]></c>.</item>
      <item>If P is <c><![CDATA[#Name.Field]]></c>, then
       Rep(P) = <c><![CDATA[{record_index,LINE,Name,Rep(Field)}]]></c>.</item>
      <item>If P is <c><![CDATA[( P_0 )]]></c>, then
       Rep(P) = <c><![CDATA[Rep(P_0)]]></c>,
       i.e., patterns cannot be distinguished from their bodies.</item>
    </list>
    <p>Note that every pattern has the same source form as some expression, and is
      represented the same way as the corresponding expression.</p>
  </section>

  <section>
    <title>Expressions</title>
    <p>A body B is a sequence of expressions <c><![CDATA[E_1, ..., E_k]]></c>, and
      Rep(B) = <c><![CDATA[[Rep(E_1), ..., Rep(E_k)]]]></c>.</p>
    <p>An expression E is one of the following alternatives:</p>
    <list type="bulleted">
      <item>If P is an atomic literal <c><![CDATA[L]]></c>, then
       Rep(P) = Rep(L).</item>
      <item>If E is <c><![CDATA[P = E_0]]></c>, then
       Rep(E) = <c><![CDATA[{match,LINE,Rep(P),Rep(E_0)}]]></c>.</item>
      <item>If E is a variable <c><![CDATA[V]]></c>, then
       Rep(E) = <c><![CDATA[{var,LINE,A}]]></c>,
       where <c><![CDATA[A]]></c> is an atom with a printname consisting of the same
       characters as <c><![CDATA[V]]></c>.</item>
      <item>If E is a tuple skeleton <c><![CDATA[{E_1, ..., E_k}]]></c>, then
       Rep(E) = <c><![CDATA[{tuple,LINE,[Rep(E_1), ..., Rep(E_k)]}]]></c>.</item>
      <item>If E is <c><![CDATA[[]]]></c>, then
       Rep(E) = <c><![CDATA[{nil,LINE}]]></c>.</item>
      <item>If E is a cons skeleton <c><![CDATA[[E_h | E_t]]]></c>, then
       Rep(E) = <c><![CDATA[{cons,LINE,Rep(E_h),Rep(E_t)}]]></c>.</item>
      <item>If E is a binary constructor <c><![CDATA[<<V_1:Size_1/TSL_1, ..., V_k:Size_k/TSL_k>>]]></c>, then
       Rep(E) = <c><![CDATA[{bin,LINE,[{bin_element,LINE,Rep(V_1),Rep(Size_1),Rep(TSL_1)}, ..., {bin_element,LINE,Rep(V_k),Rep(Size_k),Rep(TSL_k)}]}]]></c>.
       For Rep(TSL), see below.
       An omitted <c><![CDATA[Size]]></c> is represented by <c><![CDATA[default]]></c>. An omitted <c><![CDATA[TSL]]></c>
       (type specifier list) is represented by <c><![CDATA[default]]></c>.</item>
      <item>If E is <c><![CDATA[E_1 Op E_2]]></c>, where <c><![CDATA[Op]]></c> is a binary operator,
       then Rep(E) = <c><![CDATA[{op,LINE,Op,Rep(E_1),Rep(E_2)}]]></c>.</item>
      <item>If E is <c><![CDATA[Op E_0]]></c>, where <c><![CDATA[Op]]></c> is a unary operator, then
       Rep(E) = <c><![CDATA[{op,LINE,Op,Rep(E_0)}]]></c>.</item>
      <item>If E is <c><![CDATA[#Name{Field_1=E_1, ..., Field_k=E_k}]]></c>, then
       Rep(E) =
      <c><![CDATA[{record,LINE,Name,  [{record_field,LINE,Rep(Field_1),Rep(E_1)}, ..., {record_field,LINE,Rep(Field_k),Rep(E_k)}]}]]></c>.</item>
      <item>If E is <c><![CDATA[E_0#Name{Field_1=E_1, ..., Field_k=E_k}]]></c>, then
       Rep(E) =
      <c><![CDATA[{record,LINE,Rep(E_0),Name,  [{record_field,LINE,Rep(Field_1),Rep(E_1)}, ..., {record_field,LINE,Rep(Field_k),Rep(E_k)}]}]]></c>.</item>
      <item>If E is <c><![CDATA[#Name.Field]]></c>, then
       Rep(E) = <c><![CDATA[{record_index,LINE,Name,Rep(Field)}]]></c>.</item>
      <item>If E is <c><![CDATA[E_0#Name.Field]]></c>, then
       Rep(E) = <c><![CDATA[{record_field,LINE,Rep(E_0),Name,Rep(Field)}]]></c>.</item>
      <item>If E is <c><![CDATA[catch E_0]]></c>, then
       Rep(E) = <c><![CDATA[{'catch',LINE,Rep(E_0)}]]></c>.</item>
      <item>If E is <c><![CDATA[E_0(E_1, ..., E_k)]]></c>, then
       Rep(E) = <c><![CDATA[{call,LINE,Rep(E_0),[Rep(E_1), ..., Rep(E_k)]}]]></c>.</item>
      <item>If E is <c><![CDATA[E_m:E_0(E_1, ..., E_k)]]></c>, then
       Rep(E) =
      <c><![CDATA[{call,LINE,{remote,LINE,Rep(E_m),Rep(E_0)},[Rep(E_1), ...,  Rep(E_k)]}]]></c>.</item>
      <item>If E is a list comprehension <c><![CDATA[[E_0 || W_1, ..., W_k]]]></c>,
       where each <c><![CDATA[W_i]]></c> is a generator or a filter, then
       Rep(E) = <c><![CDATA[{lc,LINE,Rep(E_0),[Rep(W_1), ..., Rep(W_k)]}]]></c>. For Rep(W), see
       below.</item>
      <item>If E is a binary comprehension <c><![CDATA[<<E_0 || W_1, ..., W_k>>]]></c>,
       where each <c><![CDATA[W_i]]></c> is a generator or a filter, then
       Rep(E) = <c><![CDATA[{bc,LINE,Rep(E_0),[Rep(W_1), ..., Rep(W_k)]}]]></c>. For Rep(W), see
       below.</item>
      <item>If E is <c><![CDATA[begin B end]]></c>, where <c><![CDATA[B]]></c> is a body, then
       Rep(E) = <c><![CDATA[{block,LINE,Rep(B)}]]></c>.</item>
      <item>If E is <c><![CDATA[if Ic_1 ; ... ; Ic_k  end]]></c>,
       where each <c><![CDATA[Ic_i]]></c> is an if clause then
       Rep(E) =
      <c><![CDATA[{'if',LINE,[Rep(Ic_1), ..., Rep(Ic_k)]}]]></c>.</item>
      <item>If E is <c><![CDATA[case E_0 of Cc_1 ; ... ; Cc_k end]]></c>,
       where <c><![CDATA[E_0]]></c> is an expression and each <c><![CDATA[Cc_i]]></c> is a
       case clause then
       Rep(E) =
      <c><![CDATA[{'case',LINE,Rep(E_0),[Rep(Cc_1), ..., Rep(Cc_k)]}]]></c>.</item>
      <item>If E is <c><![CDATA[try B catch Tc_1 ; ... ; Tc_k end]]></c>,
       where <c><![CDATA[B]]></c> is a body and each <c><![CDATA[Tc_i]]></c> is a catch clause then 
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[],[Rep(Tc_1), ..., Rep(Tc_k)],[]}]]></c>.</item>
      <item>If E is <c><![CDATA[try B of Cc_1 ; ... ; Cc_k catch Tc_1 ; ... ; Tc_n end]]></c>,
       where <c><![CDATA[B]]></c> is a body, 
       each <c><![CDATA[Cc_i]]></c> is a case clause and 
       each <c><![CDATA[Tc_j]]></c> is a catch clause then 
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[Rep(Cc_1), ..., Rep(Cc_k)],[Rep(Tc_1), ..., Rep(Tc_n)],[]}]]></c>.</item>
      <item>If E is <c><![CDATA[try B after A end]]></c>,
       where <c><![CDATA[B]]></c> and <c><![CDATA[A]]></c> are bodies then
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[],[],Rep(A)}]]></c>.</item>
      <item>If E is <c><![CDATA[try B of Cc_1 ; ... ; Cc_k after A end]]></c>,
       where <c><![CDATA[B]]></c> and <c><![CDATA[A]]></c> are a bodies and
       each <c><![CDATA[Cc_i]]></c> is a case clause then 
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[Rep(Cc_1), ..., Rep(Cc_k)],[],Rep(A)}]]></c>.</item>
      <item>If E is <c><![CDATA[try B catch Tc_1 ; ... ; Tc_k after A end]]></c>,
       where <c><![CDATA[B]]></c> and <c><![CDATA[A]]></c> are bodies and
       each <c><![CDATA[Tc_i]]></c> is a catch clause then 
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[],[Rep(Tc_1), ..., Rep(Tc_k)],Rep(A)}]]></c>.</item>
      <item>If E is <c><![CDATA[try B of Cc_1 ; ... ; Cc_k  catch Tc_1 ; ... ; Tc_n after A end]]></c>,
       where <c><![CDATA[B]]></c> and <c><![CDATA[A]]></c> are a bodies, 
       each <c><![CDATA[Cc_i]]></c> is a case clause and
       each <c><![CDATA[Tc_j]]></c> is a catch clause then 
       Rep(E) =
      <c><![CDATA[{'try',LINE,Rep(B),[Rep(Cc_1), ..., Rep(Cc_k)],[Rep(Tc_1),  ..., Rep(Tc_n)],Rep(A)}]]></c>.</item>
      <item>If E is <c><![CDATA[receive Cc_1 ; ... ; Cc_k end]]></c>,
       where each <c><![CDATA[Cc_i]]></c> is a case clause then
       Rep(E) =
      <c><![CDATA[{'receive',LINE,[Rep(Cc_1), ..., Rep(Cc_k)]}]]></c>.</item>
      <item>If E is <c><![CDATA[receive Cc_1 ; ... ; Cc_k after E_0 -> B_t end]]></c>,
       where each <c><![CDATA[Cc_i]]></c> is a case clause, 
      <c><![CDATA[E_0]]></c> is an expression and <c><![CDATA[B_t]]></c> is a body, then
       Rep(E) =
      <c><![CDATA[{'receive',LINE,[Rep(Cc_1), ..., Rep(Cc_k)],Rep(E_0),Rep(B_t)}]]></c>.</item>
      <item>If E is <c><![CDATA[fun Name / Arity]]></c>, then
       Rep(E) = <c><![CDATA[{'fun',LINE,{function,Name,Arity}}]]></c>.</item>
      <item>If E is <c><![CDATA[fun Module:Name/Arity]]></c>, then
       Rep(E) = <c><![CDATA[{'fun',LINE,{function,Rep(Module),Rep(Name),Rep(Arity)}}]]></c>.
      (Before the R15 release: Rep(E) = <c><![CDATA[{'fun',LINE,{function,Module,Name,Arity}}]]></c>.)</item>
      <item>If E is <c><![CDATA[fun Fc_1 ; ... ; Fc_k end]]></c> 
       where each <c><![CDATA[Fc_i]]></c> is a function clause then Rep(E) =
      <c><![CDATA[{'fun',LINE,{clauses,[Rep(Fc_1), ..., Rep(Fc_k)]}}]]></c>.</item>
      <item>If E is <c><![CDATA[query [E_0 || W_1, ..., W_k] end]]></c>,
       where each <c><![CDATA[W_i]]></c> is a generator or a filter, then
       Rep(E) = <c><![CDATA[{'query',LINE,{lc,LINE,Rep(E_0),[Rep(W_1), ..., Rep(W_k)]}}]]></c>.
       For Rep(W), see below.</item>
      <item>If E is <c><![CDATA[E_0.Field]]></c>, a Mnesia record access
       inside a query, then
       Rep(E) = <c><![CDATA[{record_field,LINE,Rep(E_0),Rep(Field)}]]></c>.</item>
      <item>If E is <c><![CDATA[( E_0 )]]></c>, then
       Rep(E) = <c><![CDATA[Rep(E_0)]]></c>,
       i.e., parenthesized expressions cannot be distinguished from their bodies.</item>
    </list>

    <section>
      <title>Generators and filters</title>
      <p>When W is a generator or a filter (in the body of a list or binary comprehension), then:</p>
      <list type="bulleted">
        <item>If W is a generator <c><![CDATA[P <- E]]></c>, where <c><![CDATA[P]]></c> is a pattern and <c><![CDATA[E]]></c>
         is an expression, then
         Rep(W) = <c><![CDATA[{generate,LINE,Rep(P),Rep(E)}]]></c>.</item>
        <item>If W is a generator <c><![CDATA[P <= E]]></c>, where <c><![CDATA[P]]></c> is a pattern and <c><![CDATA[E]]></c>
         is an expression, then
         Rep(W) = <c><![CDATA[{b_generate,LINE,Rep(P),Rep(E)}]]></c>.</item>
        <item>If W is a filter <c><![CDATA[E]]></c>, which is an expression, then
         Rep(W) = <c><![CDATA[Rep(E)]]></c>.</item>
      </list>
    </section>

    <section>
      <title>Binary element type specifiers</title>
      <p>A type specifier list TSL for a binary element is a sequence of type
        specifiers <c><![CDATA[TS_1 - ... - TS_k]]></c>.
        Rep(TSL) = <c><![CDATA[[Rep(TS_1), ..., Rep(TS_k)]]]></c>.</p>
      <p>When TS is a type specifier for a binary element, then:</p>
      <list type="bulleted">
        <item>If TS is an atom <c><![CDATA[A]]></c>, Rep(TS) = <c><![CDATA[A]]></c>.</item>
        <item>If TS is a couple <c><![CDATA[A:Value]]></c> where <c><![CDATA[A]]></c> is an atom and <c><![CDATA[Value]]></c>
         is an integer, Rep(TS) = <c><![CDATA[{A, Value}]]></c>.</item>
      </list>
    </section>
  </section>

  <section>
    <title>Clauses</title>
    <p>There are function clauses, if clauses, case clauses 
      and catch clauses.</p>
    <p>A clause <c><![CDATA[C]]></c> is one of the following alternatives:</p>
    <list type="bulleted">
      <item>If C is a function clause <c><![CDATA[( Ps ) -> B]]></c> 
       where <c><![CDATA[Ps]]></c> is a pattern sequence and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,Rep(Ps),[],Rep(B)}]]></c>.</item>
      <item>If C is a function clause <c><![CDATA[( Ps ) when Gs -> B]]></c> 
       where <c><![CDATA[Ps]]></c> is a pattern sequence, 
      <c><![CDATA[Gs]]></c> is a guard sequence and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,Rep(Ps),Rep(Gs),Rep(B)}]]></c>.</item>
      <item>If C is an if clause <c><![CDATA[Gs -> B]]></c> 
       where <c><![CDATA[Gs]]></c> is a guard sequence and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[],Rep(Gs),Rep(B)}]]></c>.</item>
      <item>If C is a case clause <c><![CDATA[P -> B]]></c> 
       where <c><![CDATA[P]]></c> is a pattern and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep(P)],[],Rep(B)}]]></c>.</item>
      <item>If C is a case clause <c><![CDATA[P when Gs -> B]]></c> 
       where <c><![CDATA[P]]></c> is a pattern, 
      <c><![CDATA[Gs]]></c> is a guard sequence and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep(P)],Rep(Gs),Rep(B)}]]></c>.</item>
      <item>If C is a catch clause <c><![CDATA[P -> B]]></c> 
       where <c><![CDATA[P]]></c> is a pattern and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep({throw,P,_})],[],Rep(B)}]]></c>.</item>
      <item>If C is a catch clause <c><![CDATA[X : P -> B]]></c> 
       where <c><![CDATA[X]]></c> is an atomic literal or a variable pattern, 
      <c><![CDATA[P]]></c> is a pattern and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep({X,P,_})],[],Rep(B)}]]></c>.</item>
      <item>If C is a catch clause <c><![CDATA[P when Gs -> B]]></c> 
       where <c><![CDATA[P]]></c> is a pattern, <c><![CDATA[Gs]]></c> is a guard sequence 
       and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep({throw,P,_})],Rep(Gs),Rep(B)}]]></c>.</item>
      <item>If C is a catch clause <c><![CDATA[X : P when Gs -> B]]></c> 
       where <c><![CDATA[X]]></c> is an atomic literal or a variable pattern, 
      <c><![CDATA[P]]></c> is a pattern, <c><![CDATA[Gs]]></c> is a guard sequence 
       and <c><![CDATA[B]]></c> is a body, then 
       Rep(C) = <c><![CDATA[{clause,LINE,[Rep({X,P,_})],Rep(Gs),Rep(B)}]]></c>.</item>
    </list>
  </section>

  <section>
    <title>Guards</title>
    <p>A guard sequence Gs is a sequence of guards <c><![CDATA[G_1; ...; G_k]]></c>, and
      Rep(Gs) = <c><![CDATA[[Rep(G_1), ..., Rep(G_k)]]]></c>. If the guard sequence is
      empty, Rep(Gs) = <c><![CDATA[[]]]></c>.</p>
    <p>A guard G is a nonempty sequence of guard tests <c><![CDATA[Gt_1, ..., Gt_k]]></c>, and
      Rep(G) = <c><![CDATA[[Rep(Gt_1), ..., Rep(Gt_k)]]]></c>.</p>
    <p>A guard test <c><![CDATA[Gt]]></c> is one of the following alternatives:</p>
    <list type="bulleted">
      <item>If Gt is an atomic literal L, then Rep(Gt) = Rep(L).</item>
      <item>If Gt is a variable pattern <c><![CDATA[V]]></c>, then
       Rep(Gt) = <c><![CDATA[{var,LINE,A}]]></c>,
       where A is an atom with a printname consisting of the same characters as
      <c><![CDATA[V]]></c>.</item>
      <item>If Gt is a tuple skeleton <c><![CDATA[{Gt_1, ..., Gt_k}]]></c>, then
       Rep(Gt) = <c><![CDATA[{tuple,LINE,[Rep(Gt_1), ..., Rep(Gt_k)]}]]></c>.</item>
      <item>If Gt is <c><![CDATA[[]]]></c>, then
       Rep(Gt) = <c><![CDATA[{nil,LINE}]]></c>.</item>
      <item>If Gt is a cons skeleton <c><![CDATA[[Gt_h | Gt_t]]]></c>, then
       Rep(Gt) = <c><![CDATA[{cons,LINE,Rep(Gt_h),Rep(Gt_t)}]]></c>.</item>
      <item>If Gt is a binary constructor <c><![CDATA[<<Gt_1:Size_1/TSL_1, ..., Gt_k:Size_k/TSL_k>>]]></c>, then
       Rep(Gt) = <c><![CDATA[{bin,LINE,[{bin_element,LINE,Rep(Gt_1),Rep(Size_1),Rep(TSL_1)}, ..., {bin_element,LINE,Rep(Gt_k),Rep(Size_k),Rep(TSL_k)}]}]]></c>.
       For Rep(TSL), see above.
       An omitted <c><![CDATA[Size]]></c> is represented by <c><![CDATA[default]]></c>. An omitted <c><![CDATA[TSL]]></c>
       (type specifier list) is represented by <c><![CDATA[default]]></c>.</item>
      <item>If Gt is <c><![CDATA[Gt_1 Op Gt_2]]></c>, where <c><![CDATA[Op]]></c>
       is a binary operator, then Rep(Gt) = <c><![CDATA[{op,LINE,Op,Rep(Gt_1),Rep(Gt_2)}]]></c>.</item>
      <item>If Gt is <c><![CDATA[Op Gt_0]]></c>, where <c><![CDATA[Op]]></c> is a unary operator, then
       Rep(Gt) = <c><![CDATA[{op,LINE,Op,Rep(Gt_0)}]]></c>.</item>
      <item>If Gt is <c><![CDATA[#Name{Field_1=Gt_1, ..., Field_k=Gt_k}]]></c>, then
       Rep(E) =
      <c><![CDATA[{record,LINE,Name,  [{record_field,LINE,Rep(Field_1),Rep(Gt_1)}, ..., {record_field,LINE,Rep(Field_k),Rep(Gt_k)}]}]]></c>.</item>
      <item>If Gt is <c><![CDATA[#Name.Field]]></c>, then
       Rep(Gt) = <c><![CDATA[{record_index,LINE,Name,Rep(Field)}]]></c>.</item>
      <item>If Gt is <c><![CDATA[Gt_0#Name.Field]]></c>, then
       Rep(Gt) = <c><![CDATA[{record_field,LINE,Rep(Gt_0),Name,Rep(Field)}]]></c>.</item>
      <item>If Gt is <c><![CDATA[A(Gt_1, ..., Gt_k)]]></c>, where <c><![CDATA[A]]></c> is an atom, then
       Rep(Gt) = <c><![CDATA[{call,LINE,Rep(A),[Rep(Gt_1), ..., Rep(Gt_k)]}]]></c>.</item>
      <item>If Gt is <c><![CDATA[A_m:A(Gt_1, ..., Gt_k)]]></c>, where <c><![CDATA[A_m]]></c> is 
       the atom <c><![CDATA[erlang]]></c> and <c><![CDATA[A]]></c> is an atom or an operator, then
       Rep(Gt) = <c><![CDATA[{call,LINE,{remote,LINE,Rep(A_m),Rep(A)},[Rep(Gt_1), ..., Rep(Gt_k)]}]]></c>.</item>
      <item>If Gt is <c><![CDATA[{A_m,A}(Gt_1, ..., Gt_k)]]></c>, where <c><![CDATA[A_m]]></c> is 
       the atom <c><![CDATA[erlang]]></c> and <c><![CDATA[A]]></c> is an atom or an operator, then
       Rep(Gt) = <c><![CDATA[{call,LINE,Rep({A_m,A}),[Rep(Gt_1), ..., Rep(Gt_k)]}]]></c>.</item>
      <item>If Gt is <c><![CDATA[( Gt_0 )]]></c>, then
       Rep(Gt) = <c><![CDATA[Rep(Gt_0)]]></c>,
       i.e., parenthesized guard tests cannot be distinguished from their bodies.</item>
    </list>
    <p>Note that every guard test has the same source form as some expression,
      and is represented the same way as the corresponding expression.</p>
  </section>

  <section>
    <title>The abstract format after preprocessing</title>
    <p>The compilation option <c><![CDATA[debug_info]]></c> can be given to the
      compiler to have the abstract code stored in 
      the <c><![CDATA[abstract_code]]></c> chunk in the BEAM file
      (for debugging purposes).</p>
    <p>In OTP R9C and later, the <c><![CDATA[abstract_code]]></c> chunk will
      contain</p>
    <p><c><![CDATA[{raw_abstract_v1,AbstractCode}]]></c></p>
    <p>where <c><![CDATA[AbstractCode]]></c> is the abstract code as described
      in this document.</p>
    <p>In releases of OTP prior to R9C, the abstract code after some more
      processing was stored in the BEAM file. The first element of the
      tuple would be either <c><![CDATA[abstract_v1]]></c> (R7B) or <c><![CDATA[abstract_v2]]></c>
      (R8B).</p>
  </section>
</chapter>